train_fixed.py

import os
import mlflow
import mlflow.sklearn
from sklearn.ensemble import RandomForestClassifier
from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split

mlflow.set_tracking_uri("http://localhost:5000")
os.environ["MLFLOW_S3_ENDPOINT_URL"] = "http://minio-service.ml-model.svc.cluster.local:9000"
os.environ["AWS_ACCESS_KEY_ID"] = "admin"
os.environ["AWS_SECRET_ACCESS_KEY"] = "admin123"

mlflow.set_experiment("fraud-detection-experiment")

X, y = make_classification(n_samples=1000, n_features=6, random_state=42)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)

model = RandomForestClassifier(n_estimators=100, random_state=42)
model.fit(X_train, y_train)

with mlflow.start_run() as run:
    accuracy = model.score(X_test, y_test)
    mlflow.log_metric("accuracy", accuracy)
    mlflow.log_param("model_type", "RandomForestClassifier")
    mlflow.sklearn.log_model(
        sk_model=model,
        artifact_path="model",
        registered_model_name="fraud-model"
    )
    print(f"Success! Run ID: {run.info.run_id}")